Download Armadillo Is Required for Adherens Junction Assembly, Cell Polarity

Armadillo Is Required for Adherens Junction Assembly, Cell Polarity,
and Morphogenesis during Drosophila Embryogenesis
R a c h e l T. Cox,* C a t h e r i n e Kirkpatrick,* a n d M a r k Peifer**
*Curriculum in Genetics and Molecular Biology, *Department of Biology, University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina 27599-3280
Abstract. Morphological and biochemical analyses
have identified a set of proteins which together form a
structure known as the adherens junction. Elegant experiments in tissue culture support the idea that adherens junctions play a key role in cell--cell adhesion and in
organizing cells into epithelia. During normal embryonic development, cells quickly organize epithelia;
these epithelial cells participate in many of the key
morphogenetic movements of gastrulation. This
prompted the hypothesis that adherens junctions ought
to be critical for normal embryonic development.
Drosophila Armadillo, the homologue of vertebrate
13-catenin, is a core component of the adherens junction
protein complex and has been hypothesized to be es-
ELL adhesion is critically important for the development of multicellular animals as well as for the integrity of their tissues. Cells must recognize their
neighbors and respond accordingly with changes in shape
or molecular makeup. Cell--cell contacts are thought to be
particularly important during embryonic development,
when the complex morphogenetic movements of gastrulation and neurulation require cells to remain in contact, often as part of organized epithelia, while retaining the ability to move with respect to their neighbors (for review see
Gumbiner, 1992). Cell--cell contacts are not only important for normal development, but are also disrupted in certain diseases such as metastatic cancer. Tumor cells have
lost their ability to adhere normally to surrounding cells
and instead migrate to new locations in the affected organism. Some or all of these cells' adhesive junctions must be
altered or abolished; this has been recently confirmed by
molecular analysis (for reviews see Takeichi, 1993; Behtens, 1994; Birchmeier, 1995).
Epithelial tissues consist of tightly juxtaposed ceils
forming a polarized sheet with differing apical and basal
surfaces. Cell--cell adhesion between epithelial cells is me-
sential for adherens junction function in vivo. We have
used an intermediate mutant allele of armadillo,
armadilloxP33,to test these hypotheses in Drosophila
embryos. Adherens junctions cannot assemble in the
absence of Armadillo, leading to dramatic defects in
cell-cell adhesion. The epithelial ceils of the embryo
lose adhesion to each other, round up, and apparently
become mesenchymal. Mutant cells also lose their normal cell polarity. These disruptions in the integrity of
epithelia block the appropriate morphogenetic movements of gastrulation. These results provide the first
demonstration of the effect of loss of adherens junctions on Drosophila embryonic development.
Address correspondence to Mark Peifer, Department of Biology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3280. Tel.:
(919) 962-2271. Fax: (919) 962-1625. E-mail: [email protected]
diated by specific adhesive junctions, the most universal of
which is the adherens junction or zonula adherens (for review see Kemler, 1993). Adherens junctions allow cells (1)
to assemble themselves into epithelial and nonepithelial
tissues, (2) to establish and maintain cell polarity, (3) to
maintain cell shape by organizing the actin cytoskeleton,
and (4) to localize certain cell-cell signaling pathways. The
adherens junction forms a belt around the apical end of
epithelial cells; adherens junctions assembled from analogous components are found in most, if not all, multicellular animals.
Vertebrate adherens junction components include the
transmembrane protein E-cadherin, which is associated
with at least four cytoplasmic catenins: a-catenin, 13-catenin, plakoglobin, and p120 (Kemler, 1993; Reynolds et al.,
1994; Shibamoto et al., 1995; Aghib and McCrea, 1995;
Staddon et al., 1995). The fruit fly has essentially identical
adherens junctions, with homologues of E-cadherin (DEcadherin; Oda et al., 1994), ct-catenin (D-c~-catenin; Oda
et al., 1993), and 13-catenin (Armadillo; Peifer, 1993; no
plakoglobin homolog has been identified), et-Catenin, a
distant vinculin relative (Nagafuchi et al., 1991; Herrenknecht et ai., 1991), may mediate association with actin.
13-Catenin/Armadillo links cadherin and e~-catenin, and
possibly serves as a point at which complex assembly and
disassembly are regulated. Less is known about p120;
phosphorylation of this Src substrate may regulate the
© The Rockefeller University Press, 00210525/96/07/133/16 $2.00
The Journal of Cell Biology, Volume 134, Number 1, July 1996 133-148
133
C
properties of cell--cell junctions (Kinch et al., 1995). Studies using mammalian cell lines have demonstrated that
E-cadherin (Nagafuchi et al., 1987), 13-catenin (Kawanishi
et al., 1995; Oyama et al., 1994), and a-catenin (Hirano et al.,
1992) are all necessary for maintaining cell-cell adhesion.
a-Catenin was also shown to be required for assembly of
epithelia in vitro.
In addition to examining the properties of adherens
junctions in tissue culture, it is important to test their function in intact animals during normal development, in both
epithelial and nonepithelial tissues. We previously examined development of nonepithelial Drosophila ovarian
germ-line cells in the absence of Armadillo, the 13-catenin
homologue. In those cells, Armadillo and thus adherens
junctions are essential for normal cell function; in Armadillo's absence, cells lose their characteristic shapes, no
longer adhere to their neighbors, and have disorganized
actin cytoskeletons (Peifer et al., 1993). These disruptions
are consistent with the idea that adherens junctions regulate cell adhesion and anchor the actin cytoskeleton. Recently, examination of Drosophila embryos depleted for
DE-cadherin also revealed phenotypes consistent with alterations in epithelial cell organization (Uemura et al.,
1996; Tepass et al., 1996). Mice missing either E-cadherin
(Larue et al., 1994; Riethmacher et al., 1995) or 13-catenin
(Haegel et al., 1995) have been generated and the resulting
phenotypes analyzed. E-cadherin mutant embryos have severe defects in pre-implantation stages; cells of the morula
disassociate. Compaction, a process thought to require
E-cadherin, is normal, presumably due to maternally provided E-cadherin. 13-Catenin mutant embryos progress
further, possibly due to a greater maternal contribution of
13-catenin relative to E-cadherin. They implant and develop to the egg-cylinder stage. However, as gastrulation
commences, the cells of the inner cell mass lose adhesion
to each other preventing further development. These experiments provide dramatic confirmation of the role of
adherens junction components in normal embryogenesis.
However, the remaining maternal contribution of these
two proteins and the presence in mice of other cadherins
and of plakoglobin, a 13-catenin relative, complicate analysis.
This sort of analysis is simpler in Drosophila, due to (1)
the availability of techniques which allow removal of maternal gene product, revealing the earliest requirements
for a protein, (2) the possible absence of functionally redundant proteins at early stages in development, and (3)
the fact that the earliest stages of Drosophila development
occur in a nuclear syncytium which is essentially a single
cell. Since there are no other cells with which to make contact, adherens junction proteins are not essential during
this phase of development. After 13 synchronous rounds
of nuclear division unaccompanied by cytokinesis, the
6,000 nuclei cellularize synchronously to create the cellular
blastoderm. It is thus possible to generate a 6,000 cell epithelial sheet in which all cells are simultaneously deprived
of the adherens junction protein, Armadillo.
Drosophila Armadillo and vertebrate [3-catenin are
>70% identical in amino acid sequence along their entire
length (for review see Peifer, 1995). In addition to their apparent role in junctions, both Armadillo (Wieschaus et al.,
1984; Peifer et al., 1994) and f3-catenin (McCrea et al.,
1993; Heasman et al., 1994; Funayama et al., 1995) play
distinct roles in transduction of the Wingless/Wnt cell-cell
signal which mediates cell fate determination. Embryos
that are zygotically null for armadillo (arm) 1 have a segment polarity phenotype (Wieschaus et al., 1984), with alterations in the normal reiterated pattern of naked cuticle
and denticles on the larval ventral epidermis. These embryos do not, however, show any obvious defects in cell
adhesion. The large maternal contribution of wild-type
Armadillo (Wieschaus and Noell, 1986) is not sufficient
for Wingless signaling (Peifer, 1995), but it may be sufficient for adherens junction function. Previous attempts to
completely remove the maternal contribution were unsuccessful, as Armadillo function is also essential for oogenesis (Peifer et al., 1993).
To reveal the role of Armadillo in cell adhesion, we generated embryos in which the maternal and zygotic contribution consisted of protein encoded by the mutant allele
armxP33, an intermediate arm allele (Peifer and Wieschaus, 1990). In females whose germ lines are homozygous for this mutation, oogenesis occasionally succeeds,
producing embryos greatly depleted for both maternal and
zygotic Armadillo function. These embryos reveal the earliest requirements for Armadillo and allow us to test the
hypothesis that Armadillo, and by extension adherens
junctions, are essential for cell-cell adhesion and epithelial
integrity during Drosophila embryogenesis. When Armadillo is depleted from the onset of embryogenesis, adherens junctions fail to assemble, and normal cell polarity is
disrupted. As a result, striking defects in cell-cell adhesion
are observed. Rather than forming an organized epithelium, mutant cells round up and disassociate, forming a
multi-layered mesenchyme. This disruption of cell-cell adhesion provides the first demonstration that Armadillo is
required for adherens junction function during embryogenesis. Without normal cell-cell adhesion, gastrulation
and development are disrupted. These results confirm the
essential role of Armadillo in the formation of adherens
junctions and the integrity of epithelial tissues in Drosophila embryos.
The Journal of Cell Biology, Volume 134, 1996
134
Materials and Methods
Fly Stocks and Generation of Germ-line Clones
arm XP33 FRTIO1/FM7, arm rD~5 FRTIO1/FM7, and arm ns~ F R T I O 1 / F M 7
are described in Peifer and Wieschaus (1990) (embryonic phenotypes)
and Peifer et al. (1993) (germ-line clone phenotypes). To produce embryos depleted for maternal Armadillo, we produced homozygous clones
of mutant cells in the female germ line using the FLP-DFS technique of
Chou and Perrimon (1992) (FLP is a site-specific yeast recombinase, DFS
refers to the use of the dominant female sterile mutation Fs(l) o v o m ) .
The stock used to make germ-line clones, X X y f / o v o m F R T I O I / Y ; hsflpF 3 8 / h s f l p - F 3 8 is described in Chou and Perrimon (1992). crb "A22 and
crb °ss-3 are described in Tepass and Knust (1990). The DE-cadherin deficiency, D f (2R) E2, is described in Uemura et al. (1996). P[arm-cmyc] is
an autosomal insert of a fully functional arm + gene, that rescues a null arm
mutation to adult viability and fertility. The wild-type stock used was Canton S. All other mutations or chromosomes are described in Lindsley and
Zimm (1992). Germ-line clones were generated by heat-shocking second
instar larvae of the genotype arm xP33 FRTlO1/ovo m FRTIOI; hsflp-F38/+
for 3 h at 37°C as described in Peifer et al. (1993). Because of the dominant female sterile mutation, only germ cells homozygous mutant for
1. Abbreviations used in this paper: arm, armadillo; f-actin, filamentous ac-
tin; crb, crumbs.
and homozygous wild-type for 0110 D1 develop; thus, females with
ovaries have only a r m xe33 homozygous mutant germ cells,
a r m xP33
Antibody Staining, Visualization of Actin and DNA,
and Acridine Orange Staining
Eggs from females with a r m xe33 mutant germ lines were collected from
apple-juice/agar plates, washed in 0.1% Triton X-100, and dechorionated
with 50% bleach. Developing embryos were picked under halocarbon oil
and fixed for 1 h in 1:1 4% formaldehyde/0.1 M Pipes/2 mM MgSO4/1 mM
EGTA/0.1% NP-40:heptane (Peifer et al., 1993). After manual removal of
the vitelline membrane, embryos stained with anti-Arm (1:200) or anti-ctcatenin antibody (1:10) were prepared for immunofluorescence as described for ovaries in Peifer et al. (1993). For visualization of phosphotyrosine or DE-cadherin, embryos were fixed, stained, and washed in 0.1 M
phosphate buffer (pH 7.2); 0.3% Triton X-100 was added for DE-cadherin. Antibodies used were rabbit polyclonal anti-Arm (Riggleman et al.,
1990), rat monoclonal anti-a-catenin (1:10) (Oda et al., 1993), rat monoclonal anti-DE-cadherin (1:50) (Oda et al., 1994), and mouse monoclonal
anti-phosphotyrosine (UBI; 1:500). A Zeiss laser scanning confocal microscope was used to examine the stained embryos and ovaries. To visualize
actin and DNA, we followed the procedure of Orsulic and Peifer (1994).
Embryos were collected, fixed, and devitellinized as for antibody staining.
The embryos were then treated with 400 ixg/ml RNAse for 2 h at room
temperature, rinsed three times with PBS, and stained for 20 min using 10
p~g/ml propidium iodide and 5 U phalloidin (BODIPY FL phallacidin 503/
512, Molecular Probes, Eugene, OR). Embryos were washed three times
with PBS and mounted in Aquapolymount. Ovaries were prepared as described in Peifer et al. (1993). For acridine orange staining, embryos were
collected at different time points, washed, dechorionated, added to 1:1 10
~g/ml acridine orange (Sigma Chem. Co., St. Louis, MO) in 0.1 M phosphate buffer, pH 7.2:heptane and nutated for 10 min. The embryos were
transferred to a slide and covered with halocarbon oil and a coverslip. Pictures were taken immediately on either a Nikon fluorescent microscope or
a Zeiss laser scanning confocal microscope.
Scanning Electron Microscopy
Embryos were collected and dechorionated as for antibody staining, and
fixed in 1:1 25% gluteraldehyde:heptane for 20-30 min. Vitelline membranes were manually removed and the embryos were rinsed with PBS
with 1 mg/ml BSA, incubated in 2% osmium tetraoxide in PBS for 30 min,
rinsed with PBS, and dehydrated through an ethanol series (50%, 70%,
90%, 100% for 15 min each). After another rinse with 100% ethanol, critical point drying was done with C02. Dried embryos were placed on a stub
with double-stick tape and shadowed with gold palladium.
Examination of LivingEmbryos
and Cuticle Preparations
Embryos were collected and dechorionated as for antibody staining. Cellular blastoderm-stage embryos were picked under oil and placed on a
Petriperm plate (Heraeus) covered with halocarbon oil. A coverslip was
gently placed on the embryos and they were photographed periodically.
For cuticle preparations, after 2-3 d the coverslip was carefully removed
and individual embryos were immersed in 0.1% Triton X-100. The remaining oil was removed using a pulled Pasteur pipette, and the embryos
were mounted on a slide in 1:1 Hoyer's:Lactic acid and heated at 65°C for
1-3 d (Wieschaus and Niisslein-Volhard, 1986). In the crosses examining
genetic interactions between a r m and crb or the cadherin deficiency, fertilized embryos were selected under oil using the dissecting microscope,
percent hatching scored, and cuticle preparations made as above.
Preparation o f Embryo Extracts and Immunoblotting
Embryos were ground in SDS sample buffer and fractionated by SDSPAGE. Immunoblots were blocked with 5% nonfat dry milk in TBS plus
0.05% Tween-20 (TBST) and incubated with 1° and 2 ° antibodies in the
same solution. Washes were done in TBST without milk. Bound antibody
was detected by ECL (Amersham Corp., Arlington Heights, IL) as recommended and blots were stripped for reprobing as described by Amersham. Antibodies used were mouse monoclonal anti-Arm at 1:500, rat
monoclonal anti-~-catenin at 1:100 (Oda et al., 1993) and rat monoclonal
anti-DE-cadherin at 1:30 (Oda et al., 1994). For Fig. 7, living armXP3"~/
Cox et al. Armadillo a n d the Adherens Junction
a r m xP33 or armXe33/+
embryos were hand selected under halocarbon oil
based on their phenotype.
Results
Severely Reducing Armadillo Function Compromises
Embryonic Development
armadillo (arm) plays roles in both Wingless signal transduction and in adherens junctions (for review see Peifer,
1995). Zygotic null arm mutant embryos have severe segment polarity defects (Wieschaus et al., 1984; Peifer and
Wieschaus, 1990), reflecting the failure of Wingless signaling, but no prominent embryonic phenotypes resulting
from loss of adherens junction function are apparent (the
dorsal closure defect may be a subtle result of reduction in
junction function; Peifer et al., 1991). The substantial maternal contribution of Armadillo (Wieschaus and Noell,
1986) appears to be sufficient for proper junction function
but not for Wingless signal transduction. Consistent with
this, there is a substantial amount of remaining Armadillo
in arm ro3s zygotic null mutant embryos (Fig. 1 B).
The effects of eliminating Armadillo function on the
earliest stages of embryonic development can only be seen
by reducing the maternal contribution. We can generate
germ-line mosaics, i.e., heterozygous mutant females in
which all germ cells are homozygous mutant for a r m - these females thus maternally contribute no Armadillo or
only mutant Armadillo to their progeny. However, if the
germ line is homozygous for null or severe arm alleles
(lacking many of the Arm repeats making up the central
part of the protein; Fig. 1 A), oogenesis is disrupted, resulting in defects so severe that no fertilizable eggs are laid
(Wieschaus and Noell, 1986; Peifer et al., 1993). This prevents examination of mutant embryos completely lacking
maternal Armadillo.
To bypass this problem and examine the requirement
for Armadillo and adherens junctions early in embryogenesis, we used the intermediate arm allele arm xp33. This allele encodes a truncated protein (Fig. 1 A). arm xe33 protein accumulates at very low levels, even compared to
other mutant proteins; accumulation is <5% the level of
wild type (Peifer and Wieschaus, 1990; Fig. 1 C). Ovaries
in which germ-line cells are homozygous for arm xe~3 have
defects similar to but less severe than those found in ovaries with germ-line cells homozygous for the null allele
arm Y°35 (Fig. 2). The earliest defect seen in arm null mutant egg chambers is failure in adhesion between follicle
and germ cells. This defect is infrequent in arm xp33 mutant
egg chambers, with the preponderance of defects at later
stages. In arm xe33 mutant egg chambers, the nurse cell cortical actin cytoskeleton is often disorganized with abnormal inclusions of actin present (Fig. 2, F and G). Perhaps
as a result, membranes separating adjacent nurse cells are
sometimes disrupted, yielding multinucleate nurse cells
(Fig. 2, F and H). Centripetal follicle cells frequently fail to
migrate and separate the oocyte from the nurse cells (Fig.
2 E), and nurse cells often fail to transfer their contents
into the oocyte (Fig. 2 H). However, at a low frequency,
normal egg chambers are seen (Fig. 2 I). arm xe~ germ-line
mosaic females lay fertilizable eggs, unlike germ-line mosaics for the null allele, but their egg production is greatly
135
Figure 1. Introduction. (A) Wild-type Armadillo and the pro-
teins made by several mutant alleles. Indicated on the right is the
ability of the protein to function in each of Armadillo's two roles:
transduction of Wingless signal and adherens junction function.
Wild-type Armadillo consists of thirteen imperfect Arm repeats
flanked by NH2- and COOH-terminal regions. The C O O H terminus is required for transduction of Wingless signal, while Arm
repeats are important for adherens junction function (they may
also function in Wingless signaling), a r m xP33 protein is intermediate in length and degree of junction function to a r m xM19 and the
null allele a r m r°35. ( B ) Maternally contributed Armadillo can
carry out Armadillo's role in junctions. Surface view of an
arrngD35/Y embryo stained with anti-Armadillo antibody. There
is no zygotically contributed Armadillo present. The remaining
staining is residual wild-type maternal Armadillo still localizing
to the membrane at stage 13 of embryogenesis. (C) Protein made
by a r m XP33 is truncated and present at lower levels than wild type.
Embryo extract from progeny of wild-type or heterozygous
armXP33/+ animals was fractionated by S D S - P A G E and immunoblotted with anti-Armadillo antibody. The upper bands represent differentially phosphorylated forms of wild-type Armadillo.
a r m XP33 protein is present a t < 5 % the level of wild-type protein.
stages of oogenesis that are similar to but less severe than those
of the null allele. All egg chambers are stained with FITC-phalloidin (green) which stains filamentous actin and propidium iodide (red) which stains DNA. (A and B) Wild-type egg chambers.
The oocyte is posterior to the nurse cells, and nurse cells contain
single nuclei and are regularly shaped. (C and D ) a r m rD35 germline clones, showing the null phenotype. In C, nurse cells are
multinucleate (arrows) and irregularly shaped, most likely due to
defects in the actin cytoskeleton. Centripetal follicle cells (arrow)
could not migrate inward to enclose the oocyte. In D, the oocyte
is incorrectly located in the middle of the egg chamber-multinucleate nurse cells are seen, as are actin inclusions ( a r r o w h e a d ) .
( E - l ) a r m xe3s germ-line clones. Most of the defects are milder
compared to a r m rD35 germ-line clones. (E) The nurse cells often
bulge out into the oocyte; in such egg chambers the centripetal
cells will not migrate in to separate the oocyte and nurse cells. (F)
At times the cells have migrated properly, but the nurse cells are
multinucleate (arrows) and abnormal actin inclusions are seen
(arrowhead), More severe phenotypes similar to those of the
a r m YD35 embryos are present at a lower frequency. (G) The oocyte is occasionally improperly positioned in the middle of the
egg chamber. (H) Frequently, the nurse cells fail to dump their
contents into the oocyte as indicated by the large nurse cells at a
late stage in oogenesis. However, normal egg chambers are
present in small numbers and are able to undergo even sensitive
stages of oogenesis like making the network of actin filaments
which appear at the beginning of nurse cell dumping (1).
The Journal of Cell Biology, Volume 134, 1996
136
Figure 2. a r m XP33 germ-line clones have defects during later
reduced compared with wild-type. Of the eggs laid, 23%
are flat or half-filled, probably reflecting the observed
problems during oogenesis. Of the full eggs, 94% have
very lightly colored yolk and do not develop. These appear
to be unfertilized, perhaps due to defects in the anterior
micropyle, the eggshell structure that allows sperm entry.
About 6% of the full eggs are fertilized and begin embryonic development (Table I). We initially examined the
end of embryogenesis by looking at the cuticle phenotype.
Developing embryos were picked at the blastoderm stage
to eliminate unfertilized eggs, and after 24 h their cuticles
were prepared. Three distinct cuticle phenotypes were observed (Table I; Fig. 3). About half of the developing embryos arrest before secreting cuticle, leaving an empty vitelline membrane after cuticle preparation ("no cuticle"
class), 25% of the developing embryos hatch or have a
nearly normal cuticle ("wild-type" class). The remaining
25% have a more dramatic cuticle phenotype, secreting
only a few cuticle scraps with no large contiguous pieces
("crumbs-like" class). This is reminiscent of but more severe than the cuticle phenotype of crumbs class mutants;
the best characterized of these, crumbs (orb), encodes a
protein normally required for establishing and maintaining epithelial cell polarity (Tepass et al., 1990; Tepass and
Knust, 1990; Wodarz et al., 1993). Interestingly, a similar
cuticular phenotype results from depletion of maternal
and zygotic DE-cadherin (Tepass et al., 1996).
We were surprised to find three phenotypic classes in
the above experiment, since there are only two genotypes.
We thus manipulated the zygotic contribution, to determine which genotype yields which phenotype, arm is located on the X chromosome. All embryos in this experiment
have only arm xe33 contributed maternally and zygotically
from their mothers. Wild-type fathers either contribute a
Y chromosome, yielding embryos maternally and zygotically mutant for arm, or contribute a wild-type arm* gene.
Table I. Phenotypic analysis o f eggs and embryos derived from
females whose germlines are homozygous f o r arm xe33
Egg Phenotype
normal
23%
78%
E m b r y o n i c Phenotype
Unfertilized or
Make cellular
94%
6%
C u t i c l e Phenotype
arm XP33
0
~ crossed to:
arm+(ff
% "crumbs-like" % WT % "no cuticle"
28
24
48
0
46
54
0
51
49
Y
arm + . P[arm+l
Y
,
~
arm+
Y. arm+
v
re
O"
P [atm +1 = an •ut~omal
¢ol3yof• unto.spoon ean'y~ • wild-typn armadillo gane.
y-arm + . • Y ~ c ~ , n c , y2y67&, carrying • wild-tYi~ armawh'llo Sere.
Cox et al. Armadillo and the Adherens Junction
Thus, our original cross yields two possible embryonic
genotypes, armXP33/y and armXe33/+. We altered this such
that all embryos receive a zygotic arm + gene and have the
g e n o t y p e a r m X P 3 3 / + by mating germ line clone-carrying
females with males containing the arm + Y chromosome
duplication y2 Y67g, or with males homozygous for an
arm + transgene on chromosome 2 (Table I). When this is
done, the crumbs-like class disappears and the number of
wild-type embryos increases correspondingly (Table I).
The percentage of no cuticle embryos does not change;
this phenotype is thus independent of the zygotic genotype. This suggests that because of defects in oogenesis,
half of the eggs are damaged so that these embryos cannot
develop to secrete cuticle, regardless of their zygotic arm
genotype. These embryos constitute the no cuticle class.
The fact that the no cuticle class comprises almost exactly
50% of the embryos is apparently coincidental, and not
the result of Mendelian segregation. Of the embryos developing from normal eggs, embryos that receive a zygotic
arm + gene develop normally or nearly normally. In contrast, embryos that do not receive a zygotic arm + gene,
such that all maternal and zygotic protein is a r m XP33 m u tant, have the crumbs-like phenotype.
E m b r y o s with R e d u c e d L e v e l s o f A r m a d i l l o F u n c t i o n
S h o w D e f e c t s a t Gastrulation
Given the three possible endpoints of development observed above, we followed living embryos under the light
microscope to determine when the earliest disruptions in
development occurred. To correlate these observations
with the cuticle phenotypes, cuticle preparations were
made from all embryos observed. All three classes defined
by cuticle phenotype have reproducible defects during development. We supplemented these observations by staining embryos with probes that label filamentous actin (f-actin),
to outline cells and help determine cell polarity and DNA.
We also observed embryos by scanning electron microscopy.
The no cuticle class of embryos deviates from normal development before the cellular blastoderm stage (Stage 5 =
0 min.; Fig. 3). They consistently fail to cellularize their anterior ends (Figs. 3 and 4) where there is a nearly complete
absence of both cortical actin and nuclei (Fig. 4, A and B).
The nuclei excluded from the anterior third of the egg pile
up at the border between the cellularized and acellular regions, where a thick band of f-actin forms (Fig. 4, A and
C); the embryos subsequently attempt to pinch off the anterior aceUular yolk (Fig. 4 C). Once ceUularization occurs
at the egg posterior, morphogenetic movements resembling the initiation of gastrulation are sometimes observed
(Fig. 3; 30-50 min.). We suspect that the cellularization defects of the no cuticle class do not reflect normal aspects of
Armadillo function, but are a consequence of defects in
oogenesis. The anterior cytoplasm of the egg may be altered, blocking normal syncytial development. This could
be caused by defects in the anterior eggshell or yolk, due
to problems with nurse cell dumping and follicle cell migration (Fig. 2).
Embryos that make wild-type cuticles exhibit wild-type
or nearly wild-type embryogenesis. The syncytial blastoderm cellularizes normally (Fig. 3; stage 5), and cell move-
137
Living Embryos
F i g u r e 3. a r m xe33 germ-line clone-derived embryos have three cuticle phenotypes derived from three reproducible development pat-
terns. On top are shown sequential pictures of three different living embryos (anterior at left), each representing one of the possible developmental fates seen in progeny of females with a r m xm3 mutant germ lines. The inferred genotypes are indicated at the top. Embryos
of equivalent developmental stage are shown in parallel for the three classes. On the bottom are cuticle preparations of the different
classes of progeny. Wild-type embryos (on the left) have a stereotypical pattern of gastrulation and development which can be grouped
into the stages indicated. After 24 h of embryogenesis, the embryo secretes cuticle with a characteristic pattern of ventral denticle belts.
a r m xP33 germ line clone-derived embryos follow three possible paths of development. One class of embryos, shown on the far right, is
unable to cellularize around the entire egg circumference and does not develop far enough to secrete cuticle, leaving an empty vitelline
membrane. This no cuticle class attempts to constrict off the anterior part of the embryo ( a r r o w ) which cannot cellularize. The second
class of a r m xP33 mutant embryos, shown in the middle, cellularizes normally. However, at the onset of gastrulation, the cells round up
and become multilayered. Partial attempts to extend the germ band are sometimes seen. By the end of embryogenesis, this class of embryos undergoes some tissue differentiation because embryos twitch when viewed by light microscopy suggesting nerve and muscle development. In addition, balls of cells ( a r r o w ) are present which later secrete scraps of cuticle. The third class of embryos (not shown) develops like or almost like the wild-type embryos shown at the left. The wild-type stages indicated approximately the times indicated for
the mutant embryos.
The Journalof Cell Biology,Volume134, 1996
138
cells found in this mesenchyme round up (Fig. 3; 30-50
min).
a r m xe33 M u t a n t s E x h i b i t Severely D i s r u p t e d
Cell-CeU Adhesion
Figure 4. The no cuticle phenotype is made by embryos with severe defects in the organization of both f-actin and nuclei by the
onset of cellularization. (A and B) Optical confocal sections of a
no cuticle class embryo stained with FITC-phalloidin (green) to stain
f-actin and propidium iodide (red) to stain DNA. In surface view
(A), one can see that the anterior region of these embryos is unable to cellularize, perhaps due to maternal defects during oogenesis. Both nuclei and f-actin are absent in the anterior region and
the missing nuclei appear to be found just anterior to a thick band
of actin at ~30% egg length. A cross section of the same embryo
(B) shows that the cells that form become multilayered. (C) A
scanning electron micrograph of a no cuticle class mutant embryo
shows the embryo constricts where cellularization stops; the yolk
typically bulges from the anterior end. Anterior is left in all.
ments during gastrulation are either wild-type or only
marginally deviant (Fig. 3; stages 7-9). In contrast, embryos that make cuticles of the crumbs-like class (henceforth referred to as arm xp33 mutant embryos) undergo
very abnormal development. These embryos cellularize
normally (Fig. 3; 0 min), but morphogenetic movements
are disrupted at the onset of gastrulation (Fig. 3, 30-50
min). There is little or no germ band extension, and the
ventral furrow and posterior midgut fail to invaginate normally. Some embryos make partial ventral furrows and
posterior midgut invaginations that are reminiscent of normal gastrulation, while others do not make any furrows
whatsoever. In all arm xp33 mutant embryos, the epithelial
monolayer of cells at the cellular blastoderm stage rapidly
becomes a disordered, multi-layered mesenchyme; the
Cox et al. Armadillo and the Adherens Junction
Scanning electron microscopy of arm xe33 mutant embryos
reveals striking defects in cell-cell adhesion. The wild-type
cellular blastoderm consists of an epithelial monolayer of
tightly juxtaposed columnar cells (Fig. 5, A - C ) . These cells
are joined by spot adherens junctions (Tepass and Hartenstein, 1994), visible by transmission electron microscopy or
by staining with antibodies to adherens junction proteins
(see below), arm xe33 mutant embryos, which are maternally and zygotically severely depleted for Armadillo, form
a normal cellular blastoderm in which the cells are the
proper columnar shape (Fig. 5 A). During cellularization,
the developing cells of mutant embryos remain adherent
to each other. However, as soon as cellularization is completed and gastrulation begins, defects in cell shape and
organization are apparent as the cells lose adhesion to one
another. At early stages, parts of the embryo remain epithelial, while in other parts cells round up and become
multilayered (Fig. 5 D). As development proceeds, the remaining cells round up. In surface view, large, round cells
are seen that appear only loosely connected (Fig. 5 G); in
cross section many layers of round cells are seen (Fig. 5, E
and F). The same changes can be observed by examining
embryos stained with fluorescent markers for f-actin and
nuclei (Fig. 6). The polarity of f-actin accumulation is also
altered. The actin cytoskeleton of a normal ectodermal epithelial cell shows a strong polarity (Fig. 6, A, B, and/), in
which the heaviest accumulations are seen at the apical
end of the lateral surface (in particular in the adherens
junction) and along the basal surface. This polarity is rapidly lost in a r m xP33 mutant embryos as cells round up and
lose contact with each other; f-actin distribution becomes
uniform around the entire plasma membrane, with no obvious polarity (Fig. 6, C, K, and L).
By N2 h after the onset of gastrulation, while most cells
of arm Xp33 mutant embryos remain very disorganized (Fig.
5, F and I), subsets of cells recreate some types of multicellular organization (Fig. 5 I). Some cells cluster into balls
with epithelial characteristics (Fig. 5 J). In addition to fully
formed epithelial balls, one can see small groups of cells
apparently in the process of invaginating (Fig. 6 D, arrowhead). Distinctive spindle-shaped cells form another type
of multi-cellular aggregate (Fig. 5 K). Epithelial balls and
spindle-shaped cells occupy distinct positions along the anterior-posterior axis, with epithelial balls most common
anteriorly and spindle-shaped cells found medially. Each
subset may represent distinct cell types more resistant to
the disruptive effects of Armadillo depletion than their
neighbors.
A r m a d i l l o Is R e q u i r e d f o r A d h e r e n s J u n c t i o n A s s e m b l y
a n d Cell Polarity
All components of the multi-protein adherens junction
complex are thought to be essential for function. We
tested whether adherens junctions form in arm xp33 mutant
embryos, using antibodies to the known adherens junction
components DE-cadherin, ~-catenin, and Armadillo (the
139
Figure 5. a r m XP33 mutant embryos show defects in cell-cell adhesion. All panels are scanning electron micrographs. Where it was possible to determine, anterior is left and ventral is down. ( A - C ) Wild-type embryos. (A) Wild-type cellular blastoderm embryos have regular, columnar ceils. (B) Gastrulation is initiated when cells of the presumptive ventral furrow constrict their apices, and invaginate (C) to
form a tube which later becomes mesoderm. ( D - F ) a r m XP33 maternally and zygoticaUy mutant embryos. Normal gastrulation is disrupted. (D) Mutant embryos lose their columnar cells at the onset of gastrulation. At early stages, some cells (on the right) partially
maintain their columnar shapes while other cells (on the left) have rounded up and become multilayered. (E) As a r m xe33 mutant embryos get older, their cells continue to become more and more multilayered as well as eventually getting smaller (F), presumably due to
additional mitoses. (G and H) Surface views of a r m XP33 mutant (G) vs wild-type (H) embryos at equivalent stages. The surface of the
a r m XP33 embryos (G) becomes bumpy, as cells lose contact with each other. (H) In contrast, the cells in a stage 9 wild-type embryo are
tightly apposed (the round cells near the cephalic furrow are undergoing mitosis). ( I - K ) Later stages in the development of a r m XP33 m u t a n t embryos. J and K represent different embryos than that seen in I. Within about 2 h after the onset of gastrulation, a r m xe33 mutant
embryos have subsets of cells with distinct morphological characteristics that have reassembled into multi-cellular structures ( L arrows).
These cell subsets include bails of ceils that appear smooth and cohesive like epithelial cells ( a r r o w , J) and spindle-shaped cells which
extend processes (K).
truncated a r m XP33 protein reacts with a n t i - A r m a d i l l o antibodies that recognize the NH2 terminus; Peifer et al.,
1990).
W e examined b o t h the levels (Fig. 7) and the intracellular localization of junctional components (Fig. 8). Wildtype A r m a d i l l o is absent from mutants (Fig. 7 B), while
lower than wild type (Fig. 7 C). In contrast, b o t h D E - c a d herin and ot-catenin are present at levels only somewhat
reduced from wild type (Fig. 7 B). These two junctional
proteins are no m o r e r e d u c e d in levels than a cytoplasmic
protein, BicD, which was used as a control. T h e overall re-
The Journalof Cell Biology,Volume 134, 1996
140
a r m XP33 m u t a n t p r o t e i n accumulates at levels substantially
Figure 6. The cells of a r m XP33 mutant embryos become multilayered and lose actin cytoskeletal polarity once gastrulation begins. Embryos were doubly stained with phalloidin to stain f-actin ( A - D and l - L ) and propidium iodide to stain nuclei ( E - H and M - P ) . A , B, E,
F, L and M are wild-type, while the rest are arm xe33 mutant embryos. Adjacent pairs are images of the same embryo (e.g., A = E, B =
F). (A, E, I, and M) Wild-type embryos have regular, columnar cells after cellularization, with a polarized actin cytoskeleton; actin is enriched in cell-cell adherens junction at the apical-lateral interface and in cell-matrix junctions on the basal surface. (B and F) Wild-type
cells retain their columnar shape during gastrulation. (J and N) The first sign that arm xe33 mutant embryos lose cell-cell adhesion is
when cells in the cellular blastoderm lose their columnar shape, giving a bowed appearance (J) compared to wild-type (/). The corresponding nuclei, which are in a very regular array in wild-type embryos (M), slip out of place in arm xm3 mutant embryos (N). (C, G, K,
and O) Once gastrulation begins, mutant embryos quickly become multilayered (C and G) and the normal polarity of the actin cytoskeleton is lost (K and O). The cells lose their columnar shape and round up. Initially regions of the embryos remain nearly columnar next
to rounded-up cells (K). These more normal regions soon disappear. (D and H) arm xm3 mutant embryos give little or no indications of
normal gastrulation. The cells become increasingly multilayered with rounded cells stacking upon each other (L and P). In later stage
embryos (D), fully formed epithelial balls (arrow) and invaginating cells (arrowhead) are seen.
ductions in the level of protein per embryo may r e f e c t a
reduction in the number of cells in mutant embryos relative to wild type. The presence of Armadillo is thus not essential for continued stability of either DE-cadherin or
a-catenin, at least for the first few hours.
a r m xe33 mutant embryos show severely reduced levels
of Armadillo staining (Fig. 8 C). The lack of A r m accumulation in maternally and zygotically mutant embryos makes
them clearly distinguishable from siblings with a wild-type
zygotic arm + gene (Fig. 8 B). We used double-label immunofluorescence to unambiguously identify arm xe~3 mutant
embryos, allowing us to examine consequences of Armadillo's absence on the intracellular localization of DE-cadherin or a-catenin.
Removal of Armadillo resulted in dramatic alterations
in the localization of other adherens junction proteins. In
arm xP33 mutants, a-catenin no longer accumulates at the
plasma membrane, but instead is found diffusely in the cytoplasm (Fig. 8 G). a r m xe33 mutant embryos also show a
change in the intracellular distribution of DE-cadherin. In
wild-type embryos, DE-cadherin is tightly localized to the
plasma membrane and enriched in adherens junctions (Fig.
Cox et al. Armadillo and the Adherens Junction
141
Figure 7. Armadillo, DE-cadherin, and a-catenin levels are slightly reduced in a r m xe33 mutant embryos. (A) Levels of Armadillo, DEcadherin, and a-catenin in a r m xe33 mutant embryos vs heterozygous siblings. Embryos were prepared for immunofluorescence with the
indicated antibodies and viewed by confocal microscopy. The same brightness and contrast settings were used for a r m XP33 mutants and
heterozygous siblings. Overall staining is reduced for all three antigens in a r m XP33 mutants. (B) Immunoblot analysis of levels of Armadillo, DE-cadherin, and et-catenin in a r m xe33 mutants vs heterozygous siblings. We loaded equal embryo equivalents to compare protein
levels per embryo, and used BicD, a cytoplasmic protein, as a control for levels of total protein. Living embryos were hand-picked under
halocarbon oil as crumbs-class (armXP33fY) or wild-type (armXP33/+), extracts made in SDS sample buffer, and analyzed by SDS-PAGE
and sequential immunobl0ttingwith the antibodies indicated under each blot. The left two lanes contain equal embryo equivalents of either armXP33/+or armXp33/y extract, while the right two lanes are controls with different amounts of wild-type extract, armXP3S/+ embryos contain nearly as much Armadillo, DE-cadherin, and a-catenin as wild-type embryos, while armXP33fy embryos contain no wildtype Armadillo, as expected, but DE-cadherin and a-catenin levels are reduced only two- to threefold; this is roughly similar to reductions in levels of total protein, indicated by BicD. (C) Levels of wild-type Armadillo vs a r m XP~3 mutant protein (the right panel is a
shorter exposure). Wild-type, armXp33/+ or armXP33fy embryos were selected and treated as in B. Hand selection of embryos by phenotype is not perfect, as reflected by the low but detectable levels of wild-type Armadillo in embryos selected as armXP33/Y. Levels of wildtype Armadillo are roughly comparable between wild-type and armXP33/+ embryos; zygotic expression of a r m thus restores near wildtype levels even in the absence of maternal contribution, consistent with zygotic phenotypic rescue. In armXP33fY embryos, with only
arm XP33 mutant protein maternally and zygotically, levels of mutant protein are greater than seen in heterozygotes in Fig. 1 C, but mutant protein still accumulates at much lower levels than the wild-type protein in heterozygous siblings.
The Journalof Cell Biology,Volume134, 1996
142
Figure 8. Adherens junctions fail to assemble in a r m X p 3 3 mutant embryos, and normal cell polarity is lost. Embryos of the genotypes indicated at the left of each row were prepared for immunofluorescence with antibodies at the top of each column, and examined by confocal microscopy. ( A - D ) anti-Armadillo, ( E - H ) anti-a-catenin, (I-M) anti-DE-cadherin and (N-R) anti-phosphotyrosine. (A, E, L and
N) Wild-type embryos accumulate adherens junction components in a characteristic lattice pattern (A and E). In cross section (I and N),
the enrichment of these proteins in the adherens junction at the apical-lateral interface can be seen. Phosphotyrosine accumulates
heavily in adherens junctions. The polarized accumulation of DE-cadherin and phosphotyrosine serves as a marker of the polarity of the
epithelial sheet. Anti-DE-cadherin also stains the lateral sides of cells, though not as strongly (I). (B, F, J, and O). armXP33/+siblings,
with a wild-type zygotic copy of arm, show much the same pattern as wild-type with all four markers. (C, G, K, L, P, and Q) Early
a r m XP33 mutant embryos exhibit different patterns for staining than their siblings. (C) In a r m X p 3 3 mutant embryos Armadillo is greatly
reduced in level relative to their siblings; the small amount of remaining Armadillo is mostly membrane bound. (G) et-Catenin faintly
outlines cells and is brighter than anti-Armadillo staining, but there is a greatly increased amount of a-catenin in the cytoplasm. (K and
L) DE-cadherin in early arm Xe33 mutant embryos is still partially plasma membrane-associated, but there is an increased amount within
cells, as part of punctate structures which may be vesicles. DE-cadherin remaining at the membrane is no longer polarized. (P and Q)
Anti-phosphotyrosine staining shows the same loss of polarity but remains completely plasma membrane associated. (D, H, M, and R)
In later arm Xe33 mutant embryos, epithelial balls form. Their putative apical-lateral interfaces accumulate adherens junction components. Anti-Armadillo and anti--a-catenin faintly stain these epithelial balls (arrows, D and H), though overall staining has decreased
even more by this point. Anti-DE-cadherin and anti-phosphotyrosine antibodies brightly stain epithelial balls (arrows, M and R). The
staining is polarized, accumulating in a tight dot resembling the wild-type adherens junction (I and N).
Cox et al. Armadillo and the Adherens Junction
143
8/). In contrast, in arm xP33 mutant embryos DE-cadherin
accumulation at the plasma membrane is reduced, and the
overall staining intensity drops somewhat. Much of the remaining DE-cadherin accumulates within cells, presumably in the ER, Golgi, or endosomes (Fig. 8, K and L).
This may be a result of endocytosis, which occurs in cultured mammalian cells if cell adhesion is disrupted by removal of calcium (Kartenbeck et al., 1991).
Adherens junctions are also thought to be important in
the establishment of cell polarity (McNeill et al., 1990). To
examine cell polarity, we used antibodies directed against
DE-cadherin and phosphotyrosine. Phosphotyrosine accumulates in epithelial cells of D r o s o p h i l a imaginal discs in a
polarized fashion, with intense accumulation in the region
of the adherens junction (Woods and Bryant, 1993). We
have found that phosphotyrosine also serves as a polarity
marker in the embryonic ectodermal epithelium; it is
greatly enriched at the adherens junction (Fig. 8 N). In
wild-type embryos, DE-cadherin and phosphotyrosine
colocalize to adherens junctions. DE-cadherin, while apically enriched, also accumulates along the lateral cell
membranes (Fig. 8, I and N).
I n a r m Xe33 mutant embryos, cells appear to have normal
polarity during cellularization, but upon the onset of gastrulation cell polarity is rapidly disrupted. Both phosphotyrosine and DE-cadherin lose their apical enrichment and
instead are found all around the circumference of the cells
(Fig. 8, K and P). Phosphotyrosine continues to accumulate at the cell surface at roughly normal levels but in a delocalized fashion; however, it does remain enriched at
cell-cell contact surfaces.
These disruptions in adherens junctions and cell polarity
persist through the process of epithelial disintegration.
However, about 2 h after the onset of gastrulation the cells
in mutant embryos that join together to form epithelial
balls reform apparent adherens junctions (Figs. 5 and 8).
The assembly of these groups of cells can be seen using
anti-phosphotyrosine antibody (Fig. 8 R, arrow). Phosphotyrosine becomes repolarized, localizing to the inside of
epithelial balls at the apparent apical ends of the cells, as is
seen in wild-type epithelia (Fig. 8 N). A similar repolarization of DE-cadherin is seen in arm xe33 mutant embryos
(Fig. 8 M, arrow). Armadillo and t~-catenin also localize to
areas of epithelial organization (Fig. 8, D and H); their
staining is substantially fainter, yet still is enriched in putative adherens junctions. Outside epithelial balls, the situation is quite different. At this later stage in development,
many cells are labeled very poorly or not at all with all of
the antibodies.
In certain systems, when cell-cell or cell-matrix interactions are disrupted, the apoptosis program is triggered
(e.g., Hermiston and Gordon, 1995; Re et al., 1994). We
thus examined whether the dissolution of epithelia in
arm xe33 mutant embryos led to increased or premature
programmed cell death. We used acridine orange, a vital
dye, to examine the pattern and timing of cell death in
arm xm3 mutant embryos. In a wild-type embryo, programmed cell death commences at stage 11, ~ 7 h after egg
laying, and it occurs in a very reproducible spatial and
temporal pattern (Abrams et al., 1993; Fig. 9 A). In
arm xe~3 mutant embryos, there was no noticeable cell
death before 8 h (Fig. 9 C), and substantial cell death was
The Journal of Cell Biology, Volume 134, 1996
Figure 9. Programmed cell death occurs in an altered pattern in
arm xe33 mutant embryos. Acridine orange dye was used to visualize cells undergoing programmed cell death in arm xm~ mutant
embryos. (A) An optical section of an armXm3/+ embryo, show-
ing the wild-type pattern of programmed cell death- specific cells
die at specified times. Cell death normally commences at stage
11, ~7 h after egg laying. (B) While cell death in arm xp33 mutant
embryos commences at the normal time (this embryo is greater
than 12-h-old), the normal spatial pattern of programmed cell
death is not seen. Instead cells die in a disorganized, seemingly
random pattern. (C) Before 8 h after egg laying, arm xp33 mutant
embryos have no cells which are stained by the dye, suggesting
that programmed cell death has not yet begun. (A) is a laser scanning micrograph whereas both B and C are fluorescent micrographs.
not observed until after 10 h. When cell death did occur,
the spatial pattern seen in wild-type embryos was not observed, reflecting the totally altered morphology of the
mutant embryos. Instead, relatively large numbers of cells
died throughout the embryo (Fig. 9 B). This is similar to
the timing and pattern of cell death observed in c r u m b s
mutant embryos (Abrams et al., 1993).
Genetic Interactions between armadillo and crumbs Are
Additive
The similarity in cuticle phenotype between a r m XP33 m u t a n t s and mutants of the c r u m b s (crb) class (Tepass et al.,
1990; Tepass and Knust, 1990) led us to investigate genetic
interactions between arm and crb. We crossed females
heterozygous for both a null allele of arm (arm vD35) and a
strong allele of crb (crb 11A22) with males heterozygous for
crb 11A22. Among the progeny (Fig. 10 A), 3/16ths will be
hemizygous for arm rD35 and display a strong segment po-
144
larity phenotype, 3/16ths will be homozygous for crb 11A22
and secrete many small pieces of cuticle, and 1/16th will be
doubly mutant for arm Y035 and c r b 11A22. We observed the
expected arm and crb phenotypes at the expected frequencies, and also a new crb-like phenotype (Fig. 10 A) with
considerably less cuticle than c r b 11A22 homozygotes. The
frequency of embryos with this severe crb-like phenotype
suggests that they are the a r m Y D 3 5 / y ; c r b 11A22 double mutants. This cross also generates animals heterozygous for
one mutation and homozygous for the other; if any dosesensitive interactions were to be seen, they would emerge
as alterations in the phenotypes of some of the apparent
single mutants. No such dose-sensitive interactions were
observed. The observed double-mutant phenotype may
represent the additive effect of losing epithelial integrity in
an embryo which is unable to generate a full-size cuticle. A
similar cross with the weak crb °88-3 allele was also performed. Most crb 088-3 homozygotes survive embryogenesis; those which die show occasional defects such as small
cuticle holes, armY°35; crb °883 double mutants resembled
arm v°35 hemizygotes, but with occasional small holes in
the cuticle.
R e d u c i n g D E - C a d h e r i n Levels Suppresses the armadillo
S e g m e n t Polarity P h e n o t y p e
Given our data in support of the idea that the maternal
contribution of Armadillo is normally sufficient for junction function, we were curious to determine whether reduction in the levels of another junctional component
might reveal a junctional phenotype of the arm zygotic
null. We therefore generated animals hemizygous for
a r m XP33 (Fig. 10 B) and also heterozygous for a deficiency
that removes the DE-cadherin gene (Fig. 10 B, Df(2R) E2;
Uemura et al., 1996). These embryos should thus possess
half as much cadherin as normal. No increase in the severity of junctional phenotypes (e.g., cuticle integrity or dorsal closure) was noted, suggesting that in this range of dosage, the system is not dose-sensitive for junction function.
However, there was a substantial reduction in the severity
of the segment polarity phenotype (Fig. 10 B). Animals
hemizygous for arm rD35 and heterozygous for the DE-cadherin deficiency had a similar though less dramatic suppression of their segment polarity phenotype (data not
shown). We suspect that the suppression observed in both
cases reflects the fact that Armadillo's roles in adherens
junctions and Wingless signaling are separable, and that
under conditions where Armadillo is limiting, as in a
strong hypomorphic arm mutant, the reduction in the
number of junctional complexes frees up some of the wildtype maternal store of Armadillo protein, allowing it to
function in Wingless signaling. Similar competitive interactions have been seen in Xenopus, where increases in
cadherin levels can block action of 13-catenin in Wnt signaling (Heasman et al., 1994).
Figure 10. Genetic interactions between armadillo, crumbs, and
DE-cadherin. (A) To look for interactions between crb and arm,
we crossed flies doubly heterozygous for the strong alleles
crb1tA22 and arm v°35. crb11A22homozygotes produce scraps of cuticle while arm v°35 hemizygotes have a severe segment polarity
defect, and they produce a cuticle much shorter than wild-type.
arm v°3s crb11A22double mutants secrete a reduced number of cuticle scraps relative to crb Ha22 homozygotes-the two phenotypes
are roughly additive. (B) armXe33/y embryos have a severe segment polarity defect (though it is slightly less severe than that of
the null allele armY°35), armXe33/y; Df(Cad)/+ embryos, in which
the amount of cadherin has been reduced by 50%, have a segment polarity defect which is substantially less severe than
armXP33/Y alone. Genotypes indicated in some cases are inferred
from the frequency of a phenotype among the progeny.
Normal development requires that individual cells organize themselves into multi-cellular tissues (for review see
Gumbiner, 1992). The cell-cell adherens junction or
zonula adherens is thought to be a key element in the initi-
ation and maintenance of epithelial organization (for review
see Kemler, 1993). This theory of adherens junction function continues to be tested in tissue culture. However, testing the function of adherens junctions in intact animals has
proven more challenging. By using Drosophila to focus on
one adherens junction component, Armadillo/13-catenin,
we can use genetic as well as biochemical and molecular
approaches. We generated embryos essentially lacking Armadillo function and used these embryos to test both the
role of Armadillo in adherens junctions, and the role of
Cox et al. Armadillo and the Adherens Junction
145
Discussion
adherens junctions in epithelial integrity and in morphogenesis. These mutant embryos demonstrate that Armadillo is critical for the assembly and function of adherens
junctions. In the absence of Armadillo, other junctional
proteins are unable to form the required complex and cannot localize properly. As a result, the cell polarity normally present in epithelial cells is lost. Without normal adherens junctions, cells fail to form organized epithelia and
instead take on apparently mesenchymal character. The
dissolution of the blastoderm stage epithelium blocks the
normal morphological movements of gastrulation, and
thus catastrophically disrupts normal embryonic development. A similar analysis of the result of depletion of Armadillo or of the products of the bazooka and stardust
genes is reported in MOiler and Wieschaus (1996).
Armadillo and Adherens Junctions A r e Required f o r
Epithelial Integrity and Gastrulation
and never proceed to completion. This disruption is probably due in part to the loss of anchorage and polarity of the
actin cytoskeleton seen in mutant embryos (Fig. 6), confirming that these are properties conferred by the adherens junction.
One surprising feature of the development of arm xe33
mutants is that their cells later reassemble rudimentary epithelial and other multicellular structures (Figs. 5, 6, and
8). We suspect this is due to the residual Armadillo function provided by arm xe33 mutant protein, rather than being the result of an Armadillo-independent process.
arm xP33, while strong, is not null (Peifer and Wieschaus,
1990; these results). When zygotic expression of Armadillo
reaches its maximum level at the extended germ band
stage (Riggleman et al., 1989, 1990), the mutant protein
made appears to allow some cells to reassemble epithelia,
with adherens-like junctions at their apical ends. We believe that formation of epithelial balls results from residual
function rather than Armadillo-independent assembly of
junctions because mutant Armadillo protein accumulates
at the site of these junctions (Fig. 8). Relatively low levels
of Armadillo activity suffice for partial junction function,
consistent with previous observations concerning the zygotic phenotype of the known arm alleles. We found it interesting that both epithelial balls and spindle-shaped cells
arose at characteristic positions along the anterior-posterior axis. Cells must retain some information about their
intended cell fates despite the loss of epithelial integrity
and the disruptions caused by blocking gastrulation. Some
cells go on to form vestiges of normal larval structures,
such as the cuticular pieces secreted by epithelial balls.
arm xP33mutant embryos also twitch in a fashion similar to
the twitching caused by muscular contraction in wild-type
embryos, suggesting that some differentiation of muscle
and nerves may also occur.
Scientists have used cultured mammalian cells to probe
the function of adherens junction proteins (Nagafuchi et al.,
1987; Hirano et al., 1992); these data support the model in
which cadherin, ct- and 13-catenin form an integrated machine, with each component essential for function. This
model is being tested in intact animals using genetic analysis. In mice, both E-cadherin (Larue et al., 1994; Riethmacher et al., 1995) and 13-catenin (Haegel et al., 1995) are
essential for embryogenesis; mutant embryos arrest near
or soon after implantation. In nonepithelial germ cells of
Drosophila, Armadillo is required for cell architecture, adhesion, and integrity of the actin cytoskeleton. Disruptions
of these properties are observed in the phenotype of germ
cells homozygous for null or severely hypomorphic arm alleles (Peifer et al., 1993). Here we extended this analysis to
an intermediate allele, arm xe33. This somewhat less severe
allele has a similar though less severe phenotype during
oogenesis--while most mutant egg chambers arrest with
defects similar to those seen with the null allele, some
arm xe33 mutant egg chambers proceeded normally through
oogenesis, allowing us to examine embryos severely depleted for maternally contributed Armadillo.
In this report, we describe the consequences of depletion of maternally contributed Armadillo, which otherwise
partially rescues the zygotic loss of function. In Drosophila, the first thirteen nuclear cycles are unaccompanied by cytokinesis; 6,000 cells then form simultaneously.
Junctional components are dispensable in the syncytial
phase, as there is only one cell, so we can observe 6,000
cells simultaneously deprived of Armadillo. Since blastoderm cells are large, we could also examine the effects of
Armadillo depletion on intracellular localization of other
junctional proteins.
The results of this experiment are quite dramatic. As
soon as cellularization is complete and gastrulation should
begin, the normally columnar epithelial cells of the cellular
blastoderm round up, loosen their contacts with each
other and form a multi-layered mesenchyme (Fig. 5). This
disruption of the overall architecture of the embryo blocks
the complex morphogenetic movements of gastrulation.
At times one sees vestiges of normal gastrulation, such as
partial apical constriction of the ventral furrow or partial
germband elongation, but these are seen only infrequently
Biochemical evidence in vertebrates (Aberle et al., 1994;
Htilsken et al., 1994; Jou et al., 1995) and Drosophila (Pai,
L.-M., C. Kirkpatrick, and M. Peifer, unpublished data)
suggests that Armadillo links a-catenin and E-cadherin.
We found that removal of Armadillo disrupts adherens
junction assembly in embryos, dramatically altering the intracellular localization of both DE-cadherin and a-catenin
at the onset of gastrulation (Fig. 8). In the absence of Armadillo, ct-catenin is not found in adherens junctions but
instead accumulates inside the cell, with little or no enrichment at the membrane. The normal enrichment of DEcadherin in adherens junctions is lost. Levels of DE-cadherin inside cells are greatly increased, perhaps reflecting
accumulation in endocytic vesicles. Reduction in Ca 2+ levels in cultured cells causes a similar redistribution of cadherins (Kartenheck et al., 1991).
We suspect that in the absence of Armadillo, et-catenin
is no longer bound to DE-cadherin and is thus free in the
cytoplasm. This disrupts anchoring of the actin cytoskeleton at the adherens junction. DE-cadherin, stripped of its
two high affinity partners, is rapidly internalized. Steadystate levels of both a-catenin and DE-cadherin remain
near normal (Fig. 7), however and they rapidly reassemble
into apparent adherens junctions when zygotic levels of
The Journal of Cell Biology, Volume 134, 1996
146
Armadillo Is Required f o r Adherens Junction Assembly
mutant protein become sufficient to allow junction formation (Fig. 8). Relative levels of labeling suggest
that Armadillo may act in substoichiometric ratios to stimulate formation of junctions, despite the fact that in vertebrate cells, the predominant subunit of the adherens junction
is a 1:1:1 complex of E-cadherin, 13-catenin, and ct-catenin
(Ozawa and Kemler, 1992).
a r m XP33
Armadillo and Adherens Junctions Maintain Cell
Polarity Early in Development
One feature of epithelia is that the cells are polarized, with
defined apical and basolateral surfaces; lateral surfaces are
further polarized, such that specific junctional types form
in particular positions. Adherens junctions are thought to
help initiate and maintain this polarity (McNeill et al.,
1990). Depletion of functional Armadillo from embryos
disrupted cell polarity, as seen by disruptions in the polarized accumulation of phosphotyrosine (Fig. 8), normally
enriched in adherens junctions, and f-actin (Fig. 6), which
normally accumulates most heavily in adherens junctions
and at the basal surface. Both markers are polarized in
their localization in mutant embryos until the completion
of cellularization, suggesting that this initial polarization
does not require adherens junctions. Instead, it may reflect
differences in the nature of the original egg plasma membrane, which remains apical, and the newly created lateral
plasma membrane. The localization of phosphotyrosine
and f-actin is rapidly and dramatically altered in mutants
once gastrulation begins. Unlike DE-cadherin, these markers remain greatly enriched at the plasma membrane, but
become uniformly distributed around the circumference of
the cells. The cells also lose their characteristically asymmetric columnar shape and become rounded in appearance.
During later stages in development, when a subset of the
mutant cells reassemble rudimentary epithelial balls, both
polarity markers, actin and phosphotyrosine, become specifically re-polarized in this subset of cells (Figs. 6 and 8).
There is a strong correlation between reassembly of adherens junctions, as assayed by immunofluorescence to detect junctional components, and reacquisition of polarity,
as assayed by distribution of phosphotyrosine and actin.
The identity of the adherens junction-enriched protein to
which phosphotyrosine is attached is unknown. These data
suggest that it is not one of the known junctional components, as phosphotyrosine remains associated with the
plasma membrane in the absence or near absence of Armadillo, DE-cadherin, or ~t-catenin.
not until spot adherens junctions coalesce into the zonula
adherens that the fully polarized state of the epithelial
sheet is obtained. It is only at that point, for example, that
septate junctions (probable equivalents of vertebrate tight
junctions) are assembled (Tepass and Hartenstein, 1994).
From the data presented here, it is clear that Armadillo is
required for assembly of spot adherens junctions, and that
these are required for cell adhesion and for polarity at
early stages. The products of the bazooka and stardust
genes also are essential for initial assembly of adherens
junctions (Mtiller and Wieschaus, 1996)- the nature of the
products encoded by these genes is thus of great interest.
However, it is less clear how the zonula adherens is assembled and how this prompts the assembly of other junctional complexes, such as the septate junction. Certain
gene products, such as Crumbs, clearly are required for
cell polarity during later stages of embryogenesis (Tepass
et al., 1990; Tepass and Knust, 1993; Wodarz et al., 1993),
during and after assembly of the zonula adherens. However, Crumbs does not appear to be required for initial assembly of spot adherens junctions or for early cell polarity
(Grawe et al., 1996; Tepass, 1996). Thus, the development
of mature epithelial tissues is likely to require several
steps, with later cells having more elaborate machinery
than their ancestors.
We are particularly grateful to Drs. M. Takeichi and T. Uemura for generously sharing antibodies, fly stocks and unpublished data, and for helpful
discussions. We are also indebted to J. Adam for asking the questions that
resulted in the initiation of this project and for help with early experiments. We would like to thank J. Shields for keeping our lab operating
and S. Whitfield for her superb photographic help. We would also like to
thank Drs. E. Wieschaus, A. MUller, U. Tepass, and V. Hartenstein for
sharing unpublished data, Dr. U. Tepass for providing fly stocks, and Drs.
J. Carson and T. Perdue for help with scanning EM. Finally, we would like
to thank E. Cox, M. Kinch, D. Kirkpatrick, G. Maroni, M. Turnidge, and
one of the reviewers for insightful comments on the manuscript, and
members of the Peifer lab for valuable suggestions and discussions.
R. Cox was supported by National Institutes of Health (NIH) Training
grant 5T32 GM07092, and the project was supported by NIH grant
GM47857 and by a grant from the Searle Scholars Program to M. Peifer.
Received for publication 16 February 1996 and in revised form 9 April
1996.
References
We now know a great deal about the identity of the core
components of the adherens junction, the nature of their
interactions with each other, and the nature of their contributions to junction function in cultured cells and intact
animals. Many questions remain. The adherens junction is
both dynamic and complex. Morphological analysis in Drosophila demonstrates that the fully formed zonula adherens
develops only late in embryogenesis (Tepass and Hartenstein, 1994); for several hours after the onset of gastrulation, cells are joined only by spot adherens junctions. It is
Aberle, H., S. Butz, J. Stappert, H. Weissig, R. Kemler, and H. Hoschuetzky.
1994. Assembly of the cadherin-catenin complex in vitro with recombinant
proteins. Z Cell Sci. 107:3655-3663.
Abrams, J.M,, K. White, L.I. Fessler, and H. Steller. 1993. Programmed cell
death during Drosophila embryogenesis. Development. 117:29-43.
Aghib, D.F., and P.D. McCrea. 1995. The E-cadherin complex contains the Src
substrate p120. Exp. Cell Res. 218:35%369.
Behrens, J. 1994. Cell contacts, differentiation, and invasiveness of epithelial
cells. Invasion & Metastasis. 14:61-70.
Birchmeier, W. 1995. E-cadherin as a tumor (invasion) suppressor gene. Bioessays. 17:97-99.
Chou, T.-B., and N. Perrimon. 1992. Use of a yeast site-specific recombinase to
produce female germline chimeras in Drosophila. Genetics. 131:643-653.
Funayama, N., F. Fagatto, P. McCrea, and B.M. Gumbiner. 1995. Embryonic
axis induction by the Armadillo repeat domain of [3-catenin: evidence for intracellular signaling. J. Cell BioL 128:95%968.
Grawe, F., A. Wodarz, B. Lee, E. Knust, and H. Skaer. 1996. The Drosophila
genes crumbs and stardust are involved in the biogenesis of adherens junctions. Development. 122:951-959.
Gumbiner, B.M. 1992. Epithelial morphogenesis. Cell. 69:385-387.
Haegel, H., L. Larue, M. Ohsugi, L. Federov, K. Herrenknecht, and R. Kemler.
1995. Lack of 13-catenin affects mouse development at gastrulation. Development. 121:3529-3537.
Heasman, J., A. Crawford, K. Goldstone, P. Garner-Hamrick, B. Gumbiner, P.
Cox et al. Armadillo and the Adherens Junction
147
Adherens Junctions, the Zonula Adherens and the
Development of Epithelial Cells
McCrea, C. Kintner, C.Y. Noro, and C. Wylie. 1994. Overexpression of cadherins and underexpression of 13-catenin inhibit dorsal mesoderm induction
in early Xenopus embryos. Cell. 79:791-803.
Hermiston, M.L., and J.l. Gordon. 1995. In vivo analysis of cadherin function in
the mouse intestinal epithelium: essential roles in adhesion, maintenance of
differentiation, and regulation of programmed cell death. J. Cell Biol. 129:
489-506.
Herrenknecht, K., M. Ozawa, C. Eckerskorn, F. Lottspeich, and R. Kemler.
1991. The uvomorulin-anchorage protein ¢t-catenin is a vinculin homologue.
Proc. Natl. Acad. Sci. USA. 88:9156-9160.
Hirano, S., N. Kimoto, Y. Shimoyama, S. Hirohashi, and M. Takeichi, 1992.
Identification of a neural a-catenin as a key regulator of cadherin function
and multicellular organization. Cell. 70:293-301.
Htilsken, J., W. Birchmeier, and J. Behrens. 1994. E-cadherin and APC compete for the interaction with 13-catenin and the cytoskeleton. J. Cell Biol. 127:
2061-2069.
Jou, T., D.B. Stewart, J. Stappert, W.J. Nelson, and J.A. Marrs. 1995. Genetic
and biochemical dissection of protein linkages in the cadherin-catenin complex. Cell Biology. 92:5067-5071.
Kartenbeck, J., M. Schmelz, W.W, Franke, and B. Geiger. 1991. Endocytosis of
junctional cadherins in Bovine Kidney Epithelial (MDBK) cells cultured in
low Ca 2÷ ion medium. Z Cell Biol. 113:881-892.
Kawanishi, J., J. Kato, K. Sasaki, S. Fujii, N. Watanabe, and Y. Niitsu. 1995.
Loss of E-cadherin-dependent cell-cen adhesion due to mutation of the
13-catenin gene in a human cancer cell line, HSC-39. MoL Cell. Biol. 15:11751181.
Kemler, R. 1993. From cadherins to catenins: cytoplasmic protein interactions
and regulation of cell adhesion. Trends Genet. 9:317-321.
Kinch, M.S., G.J. Clark, C.J. Der, and K. Burridge. 1995. Tyrosine phosphorylation regulates the adhesions of ras-transformed breast epithelia. J. Cell BioL
130:461-471.
Larue, L., M. Ohsugi, J. Hirchenhain, and R. Kemler. 1994. E-cadherin null mutant embryos fail to form a trophectoderm epithelium. Proc. Natl. Acad. Sci.
USA. 91:8263-8267.
Lindsley, D.L., and G.G. Zimm. 1992. The genome of Drosophila melanogaster.
Academic Press, San Diego, CA.
McCrea, P.D., W.M. Brieher, and B.M. Gumbiner. 1993. Induction of a secondary body axis in Xenopus by antibodies to 13-catenin. J. Cell Biol. 123:477484.
McNeill, H., M. Ozawa, R. Kemler, and W.J. Nelson. 1990. Novel function of
the cell adhesion molecule uvomorulin as an inducer of cell surface polarity.
Cell, 62:309-316.
Miiller, H.-A.J., and E. Wieschaus. 1996. armadillo, bazooka, and stardust are
critical for formation of the zonula adherens and maintenance of the polarized blastoderm epithelium in Drosophila. J. Cell Biol. 134:149-163.
Nagafuchi, A., M. Takeichi, and S. Tsukita. 1991. The 102 kD cadherin-associated protein: similarity to vinculin and post-transcriptional regulation of expression. Cell, 65:849-857.
Nagafuchi, A., Y. Shirayoshi, K. Okazaki, K. Yasuda, and M. Takeichi. 1987.
Transformation of cell adhesion properties by exogenously introduced
E-cadherin cDNA. Nature (Lond.). 329:341-343.
Oda, H., T. Uemura, K. Shiomi, A. Nagafuchi, S. Tsukita, and M. Takeichi.
1993. Identification of a Drosophila homologue of a-catenin and its association with armadillo protein. Z Cell Biol. 121:1133-1140.
Oda, H., T. Uemura, Y. Harada, Y. Iwai, and M. Takeichi. 1994. A Drosophila
homolog of cadherin associated with Armadillo and essential for embryonic
cell-cell adhesion. Dev. Biol. 165:716-726.
Orsulic, S., and M. Peifer. 1994. A method to stain nuclei of Drosophila for confocal microscopy. Biotechniques. 16:441--447.
Oyama, T., Y. Kanai, A. Ochiai, S. Akimoto, T. Oda, K. Yanagihara, A. Nagafuchi, S. Tsukita, S. Shibamoto, F. Ito, et al. 1994. A truncated 13-catenin
disrupts the interaction between E-cadherin and a-catenin: a cause of loss of
intracellular adhesiveness in human cancer cell lines. Cancer Res. 54:62826287.
Ozawa, M., and R. Kemler. 1992. Molecular organization of the uvomorulincatenin complex. J. Cell Biol. 116:989-996.
Peifer, M. 1993. The product of the Drosophila segment polarity gene armadillo
is part of a multi-protein complex resembling the vertebrate adherens junction. J. Cell Sci. 105:993-1000.
Peifer, M. 1995. Cell adhesion and signal transduction: the Armadillo connection. Trends Cell Biol. 5:224-229.
Peifer, M., and E. Wieschaus. 1990. The segment polarity gene armadillo encodes a functionally modular protein that is the Drosophila homolog of hu-
man plakoglobin. Cell. 63:1167-1178.
Peifer, M., S. Orsulic, D. Sweeton, and E. Wieschaus. 1993. A role for the
Drosophila segment polarity gene armadillo in cell adhesion and cytoskeletal integrity during oogenesis. Development. 118:1191-1207.
Peifer, M., C. Rauskolb, M. Williams, B. Riggleman, and E. Wieschaus. 1991.
The segment polarity gene armadillo affects the wingless signaling pathway
in both embryonic and adult pattern formation. Development. 111:10281043.
Peifer, M., D. Sweeton, M. Casey, and E. Wieschaus. 1994. wingless signal and
Zeste-white 3 kinase trigger opposing changes in the intracellular distribution of Armadillo. Development. 120:369-380.
Re, F., A. Zanetti, M. Sironi, N. Polentarutti, L. Lanfrancone, E. Dejana, and F.
Colotta. 1994. Inhibition of anchorage-dependent cell spreading triggers apoptosis in cultured human endothelial cells. J. Cell Biol. 127:537-546.
Reynolds, A.B., J. Daniel, P. McCrea, M.J. Wheelock, J. Wu, and Z. Zhang.
1994. Identification of a new catenin: the tyrosine kinase substrate pl20cas
associates with E-cadherin complexes. Mol. Cell. Biol. 14:8333-8342.
Riethmacher, D., V. Brinkmann, and C. Birchmeier. 1995. A targeted mutation
in the mouse E-cadherin results in defective preimplantation development.
Genetics. 92:855-859.
Riggleman, B., P. Schedl, and E. Wieschaus. 1990. Spatial expression of the
Drosophila segment polarity gene armadillo is post-transcriptionally regulated by wingless. Cell. 63:549-560.
Riggleman, B., E. Wieschaus, and P. Schedl. 1989. Molecular analysis of the armadillo locus: uniformly distributed transcripts and a protein with novel internal repeats are associated with a Drosophila segment polarity gene.
Genes Dev. 3:96-113.
Shibamoto, S., M. Hayakawa, K. Takehchi, T. Hori, K. Miyazawa, N. Kitamura,
K.R. Johnson, M.J. Wheelock, N. Matsuyoshi, M. Takeiehi, et al. 1995. Association of p120, a tyrosine kinase substrate, with E-cadherin/catenin complexes. J. Cell Biol. 128:949-957.
Staddon, J.M., C. Smales, C. Schulze, F.S. Esch, and L.L. Rubin. 1995. p120, a
pl20-related protein (pl00), and the cadherin/catenin complex. J. Cell BioL
130:369-381.
Takeichi, M. 1993. Cadherins in cancer: implications for invasion and metastasis. Curr. Opin. Cell Biol. 5:806-811.
Tepass, U. 1996. Crumbs, a component of the apical membrane, is required for
zonula adherens formation in primary epithelia of Drosophila. Dev. Biol.
177:217-225.
Tepass, U., and E. KnusL 1990. Phenotypic and developmental analysis of mutations at the crumbs locus, a gene required for the development of epithelia
in Drosophila melanogaster Roux's Arch. Dev. Biol. 199:189-206.
Tepass, U., and E. Knust. 1993. crumbs and stardust act in a genetic pathway
that controls the organization of epithelia in Drosophila melanogaster. Dev.
Biol, 159:311-326.
Tepass, U., and V. Hartenstein. 1994. The development of cellular junctions in
the Drosophila embryo. Dev. Biol. 161:563-596.
Tepass, U., C. Theres, and E. Knust. 1990. crumbs encodes an EGF-like protein
expressed on apical membranes of Drosophila epithelia and required for organization of epithelia. Cell. 61:787-799.
Tepass, U., E. Gruszynski-DeFeo, T.A. Haaag, L. Omatyar, T, T6r6k, and V.
Hartenstein. 1996. shotgun encodes Drosophila E-cadherin and is preferentially required during cell rearrangement in the neurectoderm and other
morphogeneticany active epithelia. Genes Dev. 10:672-685.
Uemura, T., H. Oda, R. Kraut, S. Hatashi, Y. Kataoka, and M. Takeichi. 1996.
Zygotic D E-cadherin expression is required for the processes of dynamic
epithelial cell rearrangement in the Drosophila embryo. Genes Dev. 10:659671.
Wieschaus, E., and E. Noell. 1986. Specificity of embryonic lethal mutations in
Drosophila analyzed in germline clones. Roux's Arch. Dev. Biol, 195:63-73.
Wieschaus, E., and C. Ntisslein-Volhard. 1986. Looking at embryos. In Drosophila, A Practical Approach. D. B. Roberts, editor. IRL Press, Oxford, England. 199-228.
Wieschaus, E., C. Ntisslein-Volhard, and G. Jurgens. 1984. Mutations affecting
the pattern of the larval cuticle in Drosophila melanogaster: zygotic loci on
the X-chromosome and the fourth chromosome. Roux's Arch. Dev. Biol.
193:296-307.
Wodarz, A., F. Grawe, and E. Knust. 1993. Crumbs is involved in the control of
apical protein targeting during Drosophila epithelial development. Mech.
Dev. 44:175-187.
Woods, D.F., and P.J. Bryant. 1993. Apical junctions and cell signaling in epithelia. J. Cell Sci. (Suppl.). 17:171-181.
The Journal of Cell Biology, Volume 134, 1996
148